Opinion: Benchmarking in Precision Agriculture is Big Statistics

Opinion: Benchmarking in Precision Agriculture is Big Statistics

Benchmarking has become a buzzword in precision agriculture recently and before we get too excited with all of the profound insight it’s going to bestow upon us, let’s take a step back and really look at where this benchmarking stuff comes from, what it’s based on, and how it’s impacting the precision ag space.


The first step for these companies that offer benchmarking is to gather a lot of data from a lot of growers and before long they’ve got some Big Data to work with. With this Big Data pool, they can make comparisons of farms with similar hybrids planted on similar soil types. From there a grower can compare how he is yielding compared to his peers farming the same type of soil and planting similar seed.

I didn’t know what to make of this “benchmarking”, so a colleague of mine and I decided to test it out to see if there was something to this. My colleague’s dad farms a few hundred acres and collects yield data, but does nothing with it. And by nothing, I mean he doesn’t calibrate his monitor or even unload the data off his card.

This spring his dad’s card had filled up and he needed to free up some space so his monitor would continue to function. So Instead of erasing the multiple years of uncalibrated, low quality yield data we decided to use it to test out a leading company in the precision ag benchmarking space.


My colleague submitted several years’ worth of his dad’s data. A brief time later, we got a lengthy report back that showed how his dad’s farm stacked up to his neighbors in the same county and a bunch of other pages of fancy graphs which included:

• Yield distribution
• Yield by soil type
• Yield by elevation
• Cumulative rainfall and GDUs during the growing season
• Average rainfall and soil moisture by week
• Daily air temperature
• Evapotranspiration
• Soil temperature at the beginning of the season

These graphs didn’t really seem necessarily useful for specific improvements a grower could make to his operation, but they sure were fancy! The reports also went on to tell us that my colleague’s farm was below the bottom 25% of farms in our county in terms of yield. As someone who looks at a lot of yield data from our county, that did not sit well with me.

Another thing that did not sit well with me was that we didn’t get a single comment from this leading benchmarking company about the poor quality of the yield data. Take a look for yourself at one of the yield maps we submitted:

Notice not only the missing endrows but also the interesting yield ranges in the key.

Most people would be concerned about this, but as a precision ag professional, I know that benchmarking relies heavily on some really advanced data science methods I refer to as Big Statistics (BS for short) to fill in the gaps on highly questionable data like what we submitted.

We went back and looked at the data that this benchmarking company made those fancy graphs from. We determined that it was not fluff like we originally thought, but rather it was BS. So I guess one could say that benchmarking in precision ag doesn’t really work without the heavy use of BS.

In all seriousness, if I valued the idea of benchmarking and wanted to build out a benchmarking database, I would look at more than just yield and soil maps to “benchmark” my farm. In fact, step 1 might be to update the soil maps (that were drawn by hand decades ago) to something collected with sensors (like I wrote about before here) such as Veris data or even ERU layers. Next, I would collect and consider other data layers and variables besides yield such as:

• Soil test data
• As applied data for crop nutrients and lime
• Planting dates
• Seeding rates
• Crop rotation
• Tillage system
• Weather
• Slope and drainage
• Chemical programs
• Profitability (I’m told this is kinda important)

But I could see how a company could overlook those factors in their rush to the market place. Being fancy is really important in precision ag, however being first out of the gate is top dog.

But what about being accurate and effective? Nahhh, nobody cares about that, especially when your product is fancy looking AND first!

Ok, I’ll put down my cup of Haterade now and I’ll state that I’m not against innovative technology. I am just heavily in favor of good, solid science in any technology I associate myself with, no matter how impressive the smoke and mirrors are of technologies that don’t meet that standard.

What concerns me most is that this benchmarking nonsense is going to fall flat on its face once growers figure out what’s going on when it fails to deliver anything of sustenance. This only leaves a bad taste in everyone’s mouth and damages the progress that the good advanced analytic programs have made. This will make our industry take a few steps backwards at a time when we should be taking several leaps forward.

So, thank you for that, benchmarking. Thanks a lot…

And when a company that promotes advanced analytics like benchmarking as their bread and butter and then starts wholesaling chemicals and other farm-related products at rock bottom prices, you have to wonder what their real motives are. Since this arm of their business is so far out in left field from their roots in benchmarking, I’m going to have to cover what it means for the industry in another article.

Stay tuned…

Leave a Reply

Cory says:

Applause. Nailed it.

Dean Walker says:

What’s that saying….Garbage in Garbage out? Without someone overseeing and cleaning cleaning the data the results of bench marking are worse than worthless because it leads you in a false direction. Be careful who you trust to work with your data

Chuck trolley says:

I agree 100% I have seen it many times if the wrong person is entering the data.

Kim Retzlaff says:

it seems you failed to mention one important fact; roughly how many acres in his benchmarked area was available for use. Granted, these companies dont release this information but it can be relatively easy to determine if you are a precision agronomist in your area and know the Big Data company. For example, our county is the second largest county in the nation in acres for this particular company with nearly 100,000 acres of annual data. Is Benchmarking in this area significantly more valuable than in an area that only has nominal acre? Of course! Originally I had the same concern, soils maps are too general, yield data is not highly valuable unless cleaned first; but once you start looking at Mega Data, things become more clear and valuable. I think the key here is not Big Data but Mega data that is sufficient enough in your area.
By looking at your first 2 commentors, it appears that ‘head in the sand” is the only solution. “Garbage in, garbage out” only applies to totally worthless data. Your fathers data wasnt totally worthless! One must keep in mind that the people that are running these Big Data companies came from companies such as Amazon and Google and understand these concepts well, AND, their venture capitalists do also or they wouldnt be throwing 10’s of millions of dollars around without thoroughly understanding it.
But like I told one of my clients that was leary, ignore it all for the next 5-10 years without trying to understand and glean anything from it and in 5-10 years, you will just be that much farther behind than the others.

Ben D. Johnson says:

Kim, have you considered that you might be the one with your “head in the sand” about this benchmarking issue? Just because someone came from Amazon or Google it doesn’t mean they know analytics, but it might mean that they have some ideas on marketing. Check out my take on the situation here:


Kim Retzlaff says:

Furthermore, having been collecting clients yield data and developing management zones for over twenty years, the mountains of data had become useless until these Big Data companies began to appear. Are they perfect, far from it! But what better solutions are there now? One needs to start somewhere in order to refine a product rather than “throwing the baby out with the bathwater”!

Ben D. Johnson says:

I agree that big data can be useful in the right context, mainly with respect to the different factors I outlines in the article and not just 2 or 3 factors that probably don’t tell the complete story.